| In daily life,we often feel guilt,anger,anxiety,sadness,fear,dissatisfaction and other bad emotions,the bad emotions are collectively defined as negative emotions by psychologists.In negative emotions,fear is the most likely to occur.When a person’s function of affective regulation appear disturbance,especially in fear,multiple psychological illness may be induced(e.g.excessive panic,phobias,even depression and so on).Therefore,since G.Stanley.Hall et al.take up the study of fear in the 1980 s,it has become the concern of anthropology,psychology,biology,education etc.As is well-known,the video,as a dynamic,multi-channel stimulus material,can not only induce people’s emotions,but also adjust the emotions.However,through an overview of the literature,we found few people at home and abroad to analyze the effect of pleasant video on negative emotions based on physiological signals.Physiological signal,as an important medium of emotional analysis,has been demonstrated that it not only possess the objective authenticity,not easy to be susceptible to artificial manipulation,but also be able to reflect the human emotion truly.So,for above reasons,this paper proposed the analysis of pleasant video’s regulation on negative emotion through ECG signal.The specific contents and works are as follows:(1)Firstly,effective affective material are selected,including happy,fearful emotional video clips of two kinds and neutral low-awakening pictures.And then a scientific and reasonable program is designed to evaluate the selected,effective materials.Finally,we utilize independent sample t-test to analyze the scored,emotional material from the dimensions of valence and arousal.(2)In the study of pleasant video-clip’s effect on negative emotion,use the induction-modulation experiment,(i.e.,arouse the subjects’ fear by horrible video-clip,then regulate the negative emotion by pleasant video-clip).During the experiment,MP150 is used to collect the baseline,evoking and adjusting the ECG data,and establish the emotional database.(3)Before the feature extraction of physiological signals,the ECG signal is denoised by discrete wavelet transform.The filtered ECG signal is decomposed by the DB7 wavelet basis function to obtain the high frequency coefficient and the low frequency coefficient of each layer.Then,the mean,variance and energy of the first four high and low frequency coefficients are calculated(a total of 24 linear features).We extract the first-order approximation entropy,the second-order approximation entropy,the maximum Lyapunov exponent,the first-order LZ complexity,the second-order LZ complexity,the embedding dimension,the delay,the recursion rate,6th order weighted Krawtchouk moments,ect.(a total of 14 nonlinear features)(4)The relief feature selection algorithm is used to compute the weight of each feature between calm and fear,between calm and pleasure,between fear and pleasure.Then,according to the size of weight,we sort the features,selecting out the optimal combination for each emotion classification.Finally,the model of 84.7% average recognition rate is obtained by associating the optimal feature combination between fear and pleasure with SVM classifier,which establishes the basis for negative emotion adjustment analysis.(5)Macroscopically,by utilizing independent sample t-test to analyze the self-evaluated scores,we find that negative emotions of subjects gradually weakened throughout the adjustment with the pass of time,and the effect pleasant-video’s regulation on negative affect is faster than sit alone(i.e.,natural recovery).Microscopically,the optimal feature combination of the ECG signal in the adjustment stage is extracted according to step 1s within the time window n=15s,which sent into the established emotional recognition model.Then,the real-time probability curve is drawn,following the probability that the subjects’ emotional state is recognized as fear.The results show that under the mediation of pleasant-video,the average time that the subjects’ emotional state is judged to be in fear is 175.5s.Under the regulation of the natural recovery,the average time that the subjects’ emotional state is judged to be in fear is 296.5s.In terms of features,the law of change over time are as follows for each feature of the optimal feature combination between pleasure and fear:(1)Under the condition of natural recovery and pleasant-video’s regulation,the four characteristics are consistent in the changing trend.(2)Under the adjustment of pleasant video,the first-order approximate entropy’s and the maximum Lyapunov exponent’s overall change are earlier than the natural recovery regulation.Under the natural recovery regulation,the fragrant entropy’s and recursion rate’s overall change is earlier than the pleasant-video’s regulation. |
